Page 1
1
From satellite-framed Latin to verb-framed Romance:
A syntactic approach*
Víctor Acedo Matellán & Jaume Mateu
[email protected] [email protected]
(Universitat de Barcelona-Universitat Autònoma de Barcelona-CLT & Universitat Autònoma de
Barcelona-CLT)
1. Introduction
In this paper we deal with the relation between two grammatical changes from Latin to
Romance. As is well-known, the element expressing directionality –the Path– underwent a
morphological change: while the Path was typically affixed onto the verb in Latin (e.g., see
[1a]), it is typically conflated into —that is, morphologically undistinguishable from— the
verb in Romance (e.g., see the Catalan translation of [1a] in [1b]).
(1) a. E-iecit ex urbe C. Marium.
out-cast.PRF.3S out city.ABL C. Marius.ACC
‘He cast C. Marius out of the city.’
(Cic. Cat. 3, 24)1
b. Tragué Màrius de la ciutat.
cast.out.3PRF Marius of the city
However, what has gone largely unnoticed is that the morphological change exemplified in
(1) is related to syntactic changes like the one illustrated in (2), where it can be seen that
* A previous version of this paper was presented at the 10th Diacronic Generative Syntax (DIGS) conference, held at Cornell University, Ithaca, NY, in August 2008. We would like to thank the audience for useful comments and suggestions. The research carried out by Jaume Mateu has been sponsored by grants HUM2006-13295-CO2-02 (Ministerio de Ciencia e Innovación) and 2009SGR-1079 Lingüística Teòrica (Generalitat de Catalunya).
Page 2
2
while Latin allowed for unselected objects, Romance (e.g., Catalan) does not any more: in
(2a) the object putamina is not selected by the simplex intransitive verb tussire ‘to cough’, but
is possible by virtue of the presence of the directional prefix ex- ‘out’. By contrast, as
exemplified in the Catalan example of (2b), Romance does not feature unselected objects.
Accordingly, a natural translation of the unselected object construction of (2a) is given in
(2c): les closques ‘the shells’ is the direct object of a directional verb treure ‘put out’, whereas
the Latin intransitive verb tussire is translated via the adjunct d’un estossec ‘with a cough’.
(2) a. [Serpentes] putamina *(ex-)tussiunt.
snake.NOM.PL shell.ACC.PL out-cough.3PL
‘Snakes cough the egg shells out.’
(Plin. Nat. 10, 197)
b. *Les serps tussen les closques (fora).
the snakes cough.PL the shells out
c. Les serps treuen les closques d’=un estossec.
the snakes put.out.PL the shells of=a cough
‘Snakes throw out the egg shells with a cough.’
The main goal of the present paper is to show that argument structure changes like the one
exemplified in (2) are determined by the change in the morphological properties of the Path
exposed in (1). This we will carry out through the adoption of Talmy’s (1991, 2000)
typological perspective on the structure of events as interpreted through the theory of l-syntax
developed by Hale and Keyser (1993, 2002, i.a.).
Page 3
3
Our paper is structured as follows. In Section 2 we introduce Talmy’s (1991, 2000) typology
of the expression of events, structured mainly on two classes of languages: satellite-framed
languages and verb-framed languages; we also expose our account of it in syntactic terms. In
Section 3 we show that Latin belongs to the class of s-framed languages, through the
examination of some argument structure patterns in this language, and we will contrast Latin
with v-framed Romance. In Section 4 we deal with some residues of s-framedness in Old
Romance and also with a class of predicates which, seeming s-framed at first glance, can be
argued to involve a v-framed strategy when our syntactic approach is applied to them. All in
all, these facts lead us to propose that the change from the old s-framed system to the new v-
framed one was gradual. We provide overall conclusions in Section 6.
2. Talmy’s typology and l-syntax
2.1. The distinction between satellite- and verb-framed languages
In this section we provide a syntactic explanation of Talmy’s (1991, 2000) well-known
descriptive typology of s(atellite)-framed vs. v(erb)-framed languages. Before doing so, it will
be useful to introduce some relevant background on his typological work. Consider some
paradigmatic examples of his typology in (3): English can typically be taken as an example of
s-framed language, whereas Catalan can typically be regarded as an example of v-framed
language. To put it in Talmy’s (1985) terms, (3a) involves conflation of Motion with Manner,
or alternatively, in Talmy’s (1991) terms, (3a) involves conflation of MOVE with
[EVENT]SUPPORTING. By contrast, the corresponding counterpart of (3a) in a Romance
language like Catalan (see [3b]) typically involves a different lexicalization pattern, i.e.
conflation of Motion with Path, the Manner component (or the Co-event) being expressed as
an adjunct.
Page 4
4
(3) a. The bottle floated into the cave.
b. La botella entrà a la cova flotant.
the bottle entered PREP the cave floating
‘The bottle entered the cave floating.’
2.2. Conflation processes: a syntactic approach
The ‘s-framedness’ of Germanic languages like English is to be related to the fact that, for
example, the satellite directional element into in (3a) is not conflated with the verb, this null
verb being then allowed to be conflated with the so-called {“Manner constituent”/
[EVENT]SUPPORTING}. To put it in syntactic terms, the non-conflating (i.e. “satellite”) nature
of into allows the phonologically null unaccusative verb to be merged with the root √float: see
(4) for the Manner conflation pattern, which involves a syntactic compounding of a manner
root with a phonologically null event (e.g., GO); cf. Mateu & Rigau (2002, 2009), McIntyre
(2004), Zubizarreta & Oh (2007), Mateu (2008), i.a.
Page 5
5
(4) Manner conflation
v
DP v
The bottle
v PathP
v √float Path PlaceP
in-to
Place DP
in- the cave
In contrast, as can be seen in (5), the conflating nature of the Path element in Catalan yields a
directional verb entrar ‘enter’, the adjunct flotant ‘floating’ being merged outside the main
argument structure. Furthermore, as can be inferred from the English translation of the
Catalan example in (3b), a similar analysis holds for the Romance pattern involved in enter
the cave floating (cf. the Germanic pattern: float into the cave). Accordingly, the Path
lexicalization pattern is also found in (the Romance lexicon of) English.
Page 6
6
(5) Path conflation
v
DP v
La botella
v PathP
P(ath) v Path PlaceP
entrà
Place DP
a la cova
Given (5), it is then the case that v-framed languages are expected to lack complex
resultative(-like) constructions like those ones exemplified in (6), which can be identified by
two important features: on the one hand, the verb does not encode nor involve Path but pure
Manner (see Mateu & Rigau 2009)2 and, on the other, the telic PartP/PP is not an adjunct (see
Aske 1989, Mateu 2005, and Gehrke 2008, i.a.). Accordingly, all the examples in (6) involve
the Manner conflation pattern, whereby they all involve a syntactic compounding of a manner
root with a phonologically null event (i.e., GO in [6a-b] and CAUSE in [6c-g]).
Page 7
7
(6) a. John danced into the room.
b. John danced away.
c. John danced the puppet across the stage.
d. John danced the night away.
e. John danced his debts off.
f. John danced his feet sore.
g. John danced his way into a wonderful world.
3. Latin as an s-framed language
In this section, basing on Acedo-Matellán (2006a, in preparation) we will show that Latin is
an s-framed language, that is, a language where the Path component, as described above, is
morphophonologically realised as an element different from the verb. Crucially, this fact
leaves the possibility for the verb to encode another component of the event, namely manner,
through the adjunction of an independent root onto it, yielding the pattern of manner
conflation shown above. We next show constructions which arguably involve that pattern:
Unselected Object Constructions, Complex Directed Motion Constructions, Locative
Alternation constructions and, finally, constructions involving aspectual satellites.
3.1. Unselected Object Constructions
U(nselected) O(bject) C(onstruction)s are constructions whose verb, when taken out of the
construction, would not either license or theta-select the object, to use GB terminology. In (7)
through (14) there are Latin examples illustrating these two possibilities:
Page 8
8
(7) [Serpentes] putamina *(ex-)tussiunt.
snake.NOM.PL shell.ACC.PL out-cough.3PL
‘Snakes cough the egg shells out.’
(Plin. Nat. 10, 197)
(8) Omne caseum cum melle *(ab-)usus eris.
all.ACC.N.SG cheese(N)ACC.SG with honey.ABL off-use.FUT.2SG
‘Till you have use up all the cheese with honey.’
(Cat. Agr. 76, 4)
(9) Neque enim omnia emebat aut *(e-)blandiebatur.
nor in.fact everything.ACC.N.PL buy.IPFV.3SG or out-flatter.IPFV.3SG
‘Nor did he acquire his object in all cases by money or flattery.’
(Liv. 27, 31, 7)
(10) *(E-)dormi crapulam, inquam.
out-sleep.IPV.2SG intoxication.ACC.SG say.PRS.1SG
‘Sleep off that intoxication, I said.’
(Cic. Phil. 2, 30)
(11) Veniebat [...] ut sudorem illic #(ab-)lueret.
come.IPFV.3SG that sweat.ACC there off-wash.IPFV.SBJV.3SG
‘He used to go there to wash his sweat off.’
(Sen. Epist, 86, 11)
Page 9
9
(12) Purpureas=que notas filis #(in-)texuit
purple.ACC.F.PL=and motif(F)ACC.PL yarn(N)DAT.PL in-weave.PRF.3SG
albis.
white.DAT.N.PL
‘And she wove purple motifs into white yarns’
(Ov. Met. 6, 577)
(13) Haec libertus ut #(e-)bibat [...] custodis?
this.ACC.N.PL freedman.NOM that out-drink.SBJV.3SG guard.PRS.2SG
‘Are you guarding these possessions for your freedman to guzzle them all up?’
(Hor. Sat. 2, 3, 122)
(14) [Acta] quae ille in aes #(in-)cidit.
act(N)ACC.PL which.ACC.N.PL that.NOM.M.SG in brass.ACC in-cut.3SG
‘The deeds which he engraved on brass.’
(Cic. Phil. 1, 16)
Of the UOCs just presented, the ones in (7) through (10) feature otherwise intransitive verbs:
utor ‘use’ (utor takes ablative, not accusative), tussio ‘cough’, blandior ‘flatter’ (blandior
takes dative) and dormio ‘sleep’; the rest of examples feature transitive verbs, although they
do not seem to theta-select their objects: in (11) the sweat, sudorem, is not washed, but
washed off; in (12) the verb texo ‘weave’ would not select purpureas notas ‘purple motifs’, if
it did not bear a prefix; in (13) the verb bibo ‘drink’, in combination with the prefix ex-,
appears with an object which does not refer to a drinkable entity; finally, in (14) the
accusative acta ‘deeds’, is not selected by simple caedo ‘cut’, but it is possible with the
prefixed verb incido ‘cut into’, ‘engrave’.
Page 10
10
We assume a small clause approach to UOCs (Hoekstra 1988, Mateu 2001, Kratzer 2004),
where the licenser of the construction, which in the Latin examples above is the prefix, and
the object form a small clause sustaining a subject-predicate relationship, the only theta-
relationship affecting the object of the UOC. Here, furthermore, we take that licensing
predicate to be a PathP, the structure argued to be responsible for the semantics of transition.
This is exemplified in (15) through an analysis of the UOC in (9):
(15) Omnia eblandiebatur.
In the analysis of (15) the prefix starts its life as a root merged within PlaceP and then it
moves by conflation up to a phonologically null Path; from there it gets prefixed onto the verb
blandiebatur, which is the result, itself, of a manner conflation process involving the root
√bland and the light verb v. It is crucial to observe that we propose that Path gets prefixed
onto, not conflated into the verb: the Path is an element morphologically independent from
the verb, as is the case in s-framed languages. In fact, UOCs are attested in these languages,
as shown by the examples in (16) through (18):
(16) The police sirened the porsche *(to a stop).
(Clark and Clark 1979: 803)
Page 11
11
(17) Er *(ver-)gärtnerte sein gesamtes Vermögen.
he ver-garden.PST.3SG his.ACC.N.SG whole.ACC.N.SG fortune(N)ACC.SG
‘In gardening, he used up all his fortune.’
(German; Stiebels 1998: 285)
(18) Ona *(is-)pisala svoju ručku.
she out-write.PST.3SG her.ACC pen.ACC
‘She has made her pen to run out of ink through writing.’
(Russian; Spencer and Zaretskaya 1998: 17)
On the contrary, as expected, UOCs are not possible in v-framed languages like Romance
(Mateu 2001). The reason is that the Path being conflated into the verb, there is no room for
any other root to get conflated into it, and the pattern of manner conflation characteristic of
UOCs does not obtain. For instance, (19) and (20) below show that a literal rendition into
Catalan of the prefixed predicates involved in the examples of (9) and (13) is impossible:
(19) *Adular tot (fora).
flatter.INF everything out
(20) *Beure (fora) aquests béns.
drink.INF out these possessions
3.2. Complex directed motion constructions
Other constructions which are typical of s-framed languages are C(omplex) D(irected)
M(otion) C(onstruction)s: i.e., constructions which involve a telic predicate expressing a goal
of motion and, simultaneously, the manner in which that motion takes place. The Latin
examples in (21) through (25) illustrate:
Page 12
12
(21) Simulatque e navi e-gressus est dedit.
as.soon.as out ship.ABL out-walk.PRF.3SG give.PFV.3SG
‘As soon as he walked out of the ship, he handed it over.’
(Cic. Verr. 2, 2, 19)
(22) Repente ex omnibus partibus ad pabulatores ad-volaverunt.
suddenly out all.ABL.F.PL quarter(F)ABL.PL at forager.ACC.PL at-fly.PRF.3PL
‘Suddenly, they flew upon the foragers from all quarters.’
(Caes. Gall. 5, 17, 2)
(23) Qui ubi ad-equitauit portis [...] vallum intravit.
who.NOM.M.SG when at-ride.PRF.3SG door.DAT.PL wall.ACC enter.PRF.3SG
‘As soon as he rode up to the doors, he entered the camp.’
(Liv. 22, 42, 5)
(24) Navigant diebus XL ad primum emporium Indiae.
sail.PRS.3PL day.ABL.PL 40 at first.ACC.N.SG emporium(N)ACC.SG India.GEN
‘They sail up to the first emporium in India in 40 days.’
(Plin. Nat. 6, 104, 1)
(25) Repente omnes [...] in Palatium cucurrerunt.
suddenly all.NOM.PL in palace.ACC run.PRF.3PL
‘On a sudden everybody hastened to the Palace.’
(Suet. 8, 2)
In these examples the manner of motion is encoded within the verb, as gredior ‘walk’, in (21)
or equito ‘ride’, in (23), while the goal is expressed as an independent element, either a prefix
—in (21), (22) and (23)— or a PP —in (24) and (25). These predicates feature an element
Page 13
13
signalling the telic character of the predicate: a complementiser like simulatque in (21) or ubi
in (23), or an adverbial like repente ‘suddenly’ in (22) and (25), or diebus XL ‘in forty days’,
in (24). This means, in the present framework, that the element expressing goal, either a
prefix or a PP, is inside the VP, since in the absence of such an element the simplex motion
verbs gredior, volo, equito, navigo and curro yield atelic activity predicates. This is illustrated
in (26) by simplex navigo, which licenses the durative adverbial per aliquot dies ‘for some
days’:
(26) Per aliquot dies totidem=que noctes [...] navigat.
for some day.ACC.PL same.number.of=and night.ACC.PL sail.PRS.3SG
‘He sails for some days and the same number of nights.’
(Hist. Apoll. rec. B, 11)
Thus, we propose the analysis in (27) for a predicate such as the one in (24)
(27) Navigant ad emporium.
In the analysis of (27), the preposition ad is first merged as Place, since it characterises the
final location of movement; from Place it conflates up into Path to provide it with
phonological material. In cases like (21), where this element appears as a prefix, there is an
Page 14
14
additional affixation operation from Path position onto the verb. In any case, Path is not
conflated into the verb. This is what permits an independent root, in this case √navig, to
adjoin to and conflate into the null light verb. As a result, the overt expression in (24) shows a
motion verb expressing manner and a PP expressing goal. This pattern is typical of s-framed
languages, like the ones illustrated in (28) through (30):
(28) On pri-exal v Moskvu za den.
he to-drove.PRF in Moscow.ACC in day
‘He arrived in Moscow in a day.’
(Russian; Gehrke 2008: 185)
(29) Dat Jan in twee uur naar Groningen is gewandeld .
that Jan in two hour to Groningen is walked
‘That Jan has walked to Groningen in two hours.’
(Dutch; Zubizarreta and Oh 2007: 42)
(30) Jon syklet til byen på en time.
Jon biked into town in one hour
‘Jon biked into town in an hour.’
(Norwegian; Tungseth 2003: 475)
In v-framed languages like Romance, telic CDMCs are not possible, since the Path
component is conflated into v, which is no longer available for an independent manner root to
conflate into it. This is shown in (31) through a Catalan literal rendition of (23). The natural
expression in Catalan involves the use of a path verb and an adjunct expressing manner, as
shown in (32):
Page 15
15
(31) *Tan bon punt cavalcà a les portes.
as.soon.as rode.PRF.3SG at the doors
(32) Tan bon punt arribà a les portes cavalcant.
as.soon.as arrive.PRF.3SG at the doors riding
Other constructions expressing directed motion and manner are possible in v-framed
languages, namely, those whose predicates are atelic:
(33) Nedaren cap a l’illa durant/ *en mitja hora.
swim.PFV.3PL towards the=island for in half.F.SG hour.F.SG
‘They swam towards the island for half an hour.’
In these cases, however, we contend that the PP expressing goal is not within VP, but
adjoined to it, while the verb heads an unergative predicate of atelic nature:
(34) Nedaren cap a l’illa.
Italian can be said to feature prima facie exceptions to the Romance v-framed pattern, since a
few manner of motion verbs in this language, like correre ‘run’, saltare ‘jump’ or volare
‘fly’, appear to allow telic CDMCs. However, the majority of manner of motion verbs, like
camminare ‘walk’, ballare ‘dance’, nuotare ‘swim’, galleggiare ‘float’, etc., do not. The two
Page 16
16
classes of verbs are also different with respect to the choice of auxiliary for the perfect tenses:
while verbs of the former class select essere ‘be’ in telic CDMCs, verbs of the latter class do
not, as shown in (35). For two similar accounts of the contrast between correre-verbs and
camminare-verbs, see Folli and Ramchand (2005) and Mateu and Rigau (2009); see also our
footnote 2.
(35) a. Gianni è corso alla stanza (cf. Gianni ha corso [*alla stanza])
Gianni is run to.the room
‘Gianni ran to the room.’
b. *Gianni è camminato alla casa. (cf. Gianni ha camminato [*alla casa])
Gianni is walked to.the house
‘Gianni walked to the house.’
3.3. Locative alternation
The locative alternation is the possibility some verbs display of heading two different
predicates, one expressing change of location —the C(hange) O(f) L(ocation) alternant— and
the other one expressing change of state —the C(hange) O(f) S(tate) alternant. The verb
spray, in the next example, is used in both ways, respectively shown in (36a) and (36b):
(36) a. Sue sprayed paint onto the wall.
b. Sue sprayed the wall with paint.
Much as the locative alternation has been widely discussed in the literature, it has not so often
been observed that it constitutes a locus for cross-linguistic variation akin to the one involved
in CDMCs and UOCs (Mateu 2002): English, for instance, is much more permissive to this
Page 17
17
phenomenon than Romance. In fact, the example of locative alternation in (36) is impossible
in Catalan, the only use of ruixar ‘spray’, being the COS alternant in (37b):
(37) a. *Ruixar pintura sobre la paret.
spray.INF paint on the wall
b. Ruixar la paret de pintura.
spray.INF the wall of paint
According to Mateu (2002), there is a correlation between the productivity of the locative
alternation and the s-/v-framed distinction: in s-framed languages like German or English
there are more cases of locative alternation than in v-framed languages like Catalan or
Spanish. This correlation could be explained if the COL alternant in s-framed languages is
analysed as in (38): the PP onto the wall is a PathP and the verb spray is a light verb, v, with a
manner root adjoined onto it:
(38) Sue sprayed paint onto the wall.
Following the localist hypothesis (Gruber 1965; Jackendoff 1983), according to which
changes of state are construed as changes of location, we claim that the COS variant also
Page 18
18
involves a PathP (i.e., the wall changes to the state of being sprayed), but it contrasts with the
COL variant in not featuring the manner conflation pattern which has been shown above; it is,
thus, a v-framed construction. Another crucial difference is that the specifier of PathP,
referring to the entity undergoing the change of state is the wall in the COS alternant:
(39) Sue sprayed the wall (with paint).
In conformity with our hypothesis that Latin is an s-framed language and that the productivity
of the locative alternation is sensitive to this typological distinction, the prediction emerges
that in Latin that productivity should be similar to the one obtaining in s-framed languages
like Germanic or Slavic, and higher than in Romance languages like Catalan. For example,
the verbs in (40) through (42) only allow the COS variant (exemplified by the b-sentences) in
Catalan:3
(40) Farcio ‘stuff’
a. In os farciri pannos imperavit.
in mouth.ACC stuff.INF.PASS rag.ACC.PL order.PFV.3SG
‘He ordered to stuff the rags into his mouth.’
(Sen. Dial. 3-4-5, 3, 19, 4)
Page 19
19
b. Medios parietes farcire fractis caementis.
central.ACC.M.PL wall(M)ACC.PL stuff.INF broken.ABL.N.PL quarry-stone(N)ABL.PL
‘To stuff the central part of a wall with fragments of quarry-stones.’
(Plin. Nat. 36, 172, 5)
(41) Lino ‘smear’
a. Medicamina [...] ut [...] lini per corpora possint.
makeup.NOM .PL that smear.INF.PASS through body.ACC.PL can.SBJV.3PL
‘Such a makeup as may be smeared on the body.’
(Ov. Medic. 81)
b. Vipereo spicula felle linunt.
of.viper.ABL.N.SG arrow(N)ACC.PL bile(N)ABL.SG smear.PRS.3PL
‘They smear all their arrows with viper bile.’
(Ov. Pont. 1, 2, 16)
(42) Stipo ‘cram’
a. Asses [...] in aliqua cella stipabant.
coin.ACC.PL in some.ABL.F.SG room(F)ABL.SG cram.IPFV.3PL
‘They used to cram the coins in some room.’
(Varro, Ling. 5, 36)
b. Senatum stiparit armatis.
senate.ACC cram.PRF.SBJV.3SG armed.ABL.M.PL
‘(That) he had crammed the senate with armed men.’
(Cic. Phil. 3, 30)
The examples just shown conform to the English pattern of (36) in that the verb of neither
alternant (farcio, lino, stipo) shows any special morphological mark. However, we do find
Page 20
20
alternations in Latin, analogous to alternations in other languages, where some alternant or
both is signalled with some morphological mark. Thus, for instance, according to Lemaire
(1983) (and confirmed in Gaffiot [1934]), the prefixed verb adfundo in (43a), ‘throw onto’,
which is found in COL expressions, is not found in COS alternants, where the verb fundo
needs another prefix, namely per ‘through’, ‘in all the extension of’, as shown in (43b):
(43) a. Ad [...] radices amphoram [...] amurcae [...] ad-fundi.
at root.ACC.PL amphora.ACC oil.residue.GEN at-throw.INF.PASS
‘For an amphora of oil residue to be thrown at the roots.’
(Plin. Nat. 17, 263, 3)
b. Caput tamen aqua frigida per-fundere.
head.ACC nonetheless water(F)ABL.SG cold.ABL.F.SG through-throw.INF
‘Wash the head, nonetheless, with cold water.’
(Cels. 1, 4)
This prefix of completive meaning is, in fact, found in other different predicates, like the one
in (44):
(44) Ausculta porro, dum hoc [...] per-lego.
listen.IPV.2SG forth until this.ACC.N.SG through-read.1SG
‘Keep listening until I have read this over.’
(Plaut. Bacch. 1005)
The pattern of (43) is analogous to the one we find in Hungarian, another s-framed language
(Snyder 2001). Ackerman (1992: 59) shows that COL variants in Hungarian are typically
Page 21
21
marked with a directional element (ra- ‘on’, in [45a]), and COS variants with a
grammaticalised element endowed with perfective (meg-, in [45b]) or resultative (tele-, in
[45c]) meaning:
(45) a. A paraszt (rá-)rakta a szénát a szekérre.
the peasant on-load.PST.3SG the hay.ACC the cart.SUBLAT
‘The peasant loaded the hay onto the cart.’
b. A cparaszt meg-rakta a szekeret (szénával).
the peasant PFV-load.PST.3SG the cart.ACC hay.INS
‘The peasant loaded the cart with hay.’
c. A paraszt tele-rakta a szekeret (szénával).
the peasant RES-load.PST.3SG the cart.ACC hay.INS
‘The peasant loaded the cart with hay.’
Importantly, Ackerman (1992) shows that neither meg- nor tele- may appear in a COL
pattern, in a way parallel to Latin perfundere. We hypothesise that the Latin and Hungarian
prefixed COS alternants represented by Latin perfundere in (43b) or by Hungarian megrakta
and telerakta in (45b) and (45c), respectively, are s-framed constructions, with the prefix
initially merged as a predicative piece indicating complete affection, and the root being
conflated with a manner root. Thus, the analysis of (43b) would be as shown in (46):
Page 22
22
(46) Caput (aqua frigida) perfundere.
The interpretation of these grammaticalised prefixes as predicates is also Mulder’s (1992:
179-180) claim for COS variants headed by be-verbs in Dutch as in (47b). That be- is to be
analysed as a predicate is suggested by the fact that it can be shown to be in complementary
distribution with adjectives like vol ‘full’ when used as resultative predicates, as shown in
(47d). (47a) represents the COL alternant and (47c) is an adjectival resultative construction
semantically near to the be-alternant in (47b):
(47) a. Hij hangt foto’s op de muur.
he hangs photos on the wall
‘He hangs photos on the wall.’
b. Hij be-hangt de muur met foto’s.
he be-hangs the wall with photos
c. Hij hangt de muur vol me foto’s.
He hangs the wall full with photos
d. *Hij be-hangt de muur vol me foto’s.
he be-hangs the wall full with photos
Page 23
23
Note that the analysis in (46) resembles the one proposed by Mulder (1992) for COS
alternants, as shown in (48):
(48) hij hangt [SC de muur {be-/vol}]
Yet a last type of alternation that we find in Latin is the one where both alternants are marked
with a directional prefix, as illustrated in (49):
(49) a. Ulceri medicamentum [...] in-ducatur.
ulcer(N)DAT.SG medicament.ABL.N.SG in-lead.PRS.SBJV.PASS.3SG
‘The medicament be smeared into the ulcer.’
(Cels. 7, 7)
b. Postes [...] sunt in-ducti pice.
door-post.NOM.PL be.PRS.3PL in-lead.PTCP.PST.NOM.M.PL pitch.ABL.SG
‘The door-posts have been smeared with pitch.’
(Plaut. Most. 827)
Our hypothesis is that the COS variants, such as the one in (49b), are s-framed constructions
with the same structure as the COL alternants: the prefix in these constructions also signals
the Path as realised independently from the verb. The difference, however is that these
constructions lack a Figure. They would be akin to the English constructions in (50), provided
by McIntyre (2004), who claims that their objects are conceptual Grounds:4
Page 24
24
(50) Wipe the table off (= wipe dust off the table), pour the bucket out (= pour water out of the
bucket), squeeze the orange out, pump the cellar out, run someone through
(McIntyre 2004: 538)
We propose the following structure for the predicate in (49b):
(51) Postes sunt inducti (pice).
Drawing on Svenonius’s (2003) and Mateu’s (2008) analyses of equivalent predicates in
English and German, respectively, we propose that the Ground (postes), generated in the
complement of Place, ends up being the accusative object, as it cannot get case from Place
(in-); thus, these predicates can be claimed to involve a kind of Burzio effect, where the
absence of the argument that would occupy the specifier/subject position of the inner SC-like
PP (i.e., the Figure) is related to the incapability of Place to assign case to the Ground.
To sum up, the s-framed pattern is also at work in the locative alternation, always in COL
variants and sometimes in COS alternants —in particular, in those marked with either a
directional or a grammaticalised prefix. This is further evidence of the preferably s-framed
nature of Latin in contrast to its daughter languages.
Page 25
25
3.5. Aspectual satellites
Satellites as defined by Talmy (2000) may not only convey spatial notions, but also aspectual
ones. According to this linguist, the typological distinction between s- and v-framed
languages can also be drawn with respect to the expression of aspect: s-framed languages are
allowed to realise aspect in the satellite, while v-framed languages can only realise it in the
verb. This contrast is shown in (52) and (53) with English and Catalan as sample languages:
(52) She played (*her guitar) on
(McIntyre 2004: 529)
(53) Ella continuà tocant (la guitarra)
she kept playing the guitar
In (52) the satellite on expresses continuous aspect, while in (53) it is the verb continuar
‘keep’, which expresses that aspectual notion. In Catalan, there is no other way to express
continuous aspect other than using an aspectual verb. It is also worth noting that, as pointed
out by McIntyre (2004), on makes the verb atransitive, that is, unable to link an argument
(her guitar, in this case). This interference of the particle on with the argument structure of
the predicate suggests that it should be located low in the structure, within the vP, as shown in
(54) (cf. She went on playing):
Page 26
26
(54) She played on.
On seems to be semantically quite similar to the Latin independent particle porro ‘forth’,
exemplified in (55) and (56):
(55) Quid fit deinde? porro loquere.
what.NOM.N.SG be.done.PRS.3SG then forth talk.IPV.2SG
‘What happened afterwards? Talk on.’
(Plaut. Amph. 1119)
(56) Ausculta porro.
listen.IPV.2SG forth
‘Keep listening.’
(Plaut. Bacch. 1005)
However, porro, unlike English aspectual on, does not seem to interfere with the argument
structure of the predicate. Thus, for instance, a verb like ausculto may appear as intransitive,
as in (56), or as transitive, as in (57), independently of the fact that porro is present in both
predicates:
Page 27
27
(57) Porro etiam ausculta pugnam.
forth even listen.IPV.2SG fight.ACC
‘Keep listening and learn about the dirty trick.’
(Plaut. Bacch. 273)
In this section we will not concentrate on this non-trivial difference between English on and
Latin porro, acknowledging that it deserves a more profound analysis, and we will limit
ourselves to showing how aspect is encoded in satellites in Latin.
The other satellites with an aspectual nature we find in Latin are prefixes. They may bear
ingressive aspect, that is, they may indicate the beginning of an event, as the prefix ad- in
(58):5
(58) Ac-cidunt arbores, tantum ut summa species
at-cut.3PL tree.ACC.PL only that highest.NOM.F.SG appearance(F)NOM.SG
earum stantium relinquatur.
this.GEN.F.PL stand.PTCP.PRS.GEN.F.PL remain.PRS.SBJV.3SG
‘They hew into the trees just to leave the appearance, on top, that they are upright.’
(Caes. Gall. 6, 27, 4)
Ad- seems quite similar in meaning to the German satellite particle an- in the prefixed verbs
of (59), taken from McIntyre (2002: 104):
Page 28
28
(59) anbraten ‘roast partly’, anbrennen ‘burn partly’, anknabbern ‘nibble partly’, ankratzen
‘scratch lightly’, andiskutieren ‘begin discussing’, antrocknen ‘dry partly’, ansägen ‘cut
partly with a saw’.
Importantly, neither ad- nor an- induce changes in argument structure properties, as opposed
to English progressive on. This could mean that the Latin and the German prefixes are to be
treated in a different way.
Prefixes may also express egressive aspect, that is, the end of the event, as the prefix de-in
(60):
(60) Dum musteus fructus de-fervescat.
until sweet.NOM.M.SG fruit(M)NOM.SG down-boil.PRS.SBJV.3SG
‘Until the sweet fruit has stopped fermenting.’
(Colum. 9, 15)
Talmy (2000: 122) provides a prefixed predicate in Russian unsurprisingly similar to the
Latin example involving de- in (60); it is shown in (61):
(61) Pivo pere-brodilo.
beer pere-fermented
‘The beer has finished fermenting.’
Another aspectual notion we find conveyed by a prefix is perfective aspect; the prefix per-
induces perfective aspect, as shown in (62):
Page 29
29
(62) Ausculta porro, dum hoc [...] per-lego.
listen.IPV.2SG forth until this.ACC.N.SG through-read.1SG
‘Keep listening until I have read this over.’
(Plaut. Bacch. 1005)
Thus, per- could be compared to particles through or over in English.
Finally, prefixes may express reciprocation. An example of this possibility is in (63):
(63) Cum risi, ad-rides.
when smile.PRF.1SG at-smile.2SG
‘When I smiled, you smiled back.’
(Ov. Met. 3, 455)
This use of ad- is similar to the one shown by English back, as the translation shows.
To conclude, the expression of aspect seems to be sensitive to the s-/v-framed distinction, s-
framed languages being allowed to express aspect in the shape of a satellite. The existence of
aspectual satellites, the wide availability of the locative alternation, the existence of UOCs
and the possibility of manner of motion verbs to appear in telic motion predicates are proofs
that Latin is an s-framed language, as initially suggested by Talmy (1985: 103). Crucially,
what all these patterns have in common is the expression of the element conveying change
(the Path) as morphophonologically independent from the verb, although it is usually prefixed
onto it, and the expression of the manner component in the verb.
Page 30
30
4. The path from s-framed Latin to v-framed Romance
Contrasting with Latin, in modern Romance, particularly in modern Catalan, the Path
component is typically no further distinguishable from the verb. This morphophonological
change brought about important consequences in the domain of argument structure in the new
system, specifically the unavailability of the manner conflation pattern we introduced in
Section 2: since the null verb v ends up hosting, by conflation, the phonological matrix
coming from Path, it can no longer accept a manner root adjoined onto it, as we suggested
may happen in s-framed constructions. In this section we will concentrate on two phenomena
involved in the lexicon of Romance which illustrate the change from the old s-framed system
into the new v-framed system. We will show, first, how the older stages of Romance still
present an s-framed pattern with manner conflation; this will be illustrated through data from
Old Catalan and Old French. Second, we will concentrate on a class of predicates which have
been claimed to represent an s-framed pattern, namely the group of Romance prefixed verbs
of change of state/location. We will take pains to show that these predicates are in reality v-
framed constructions, but ones where the Place component is specified by the prefix. Both
phenomena illustrate the gradual change from Latin to Romance within the domain of
argument structure.
4.1. Latin s-framedness in Early Romance: Old Catalan and Old French
Residues of the s-framed system involving the manner conflation mechanism are traceable in
Old Romance. Here we will pay attention, first, to the role of verbal prefixes as aspectual
markers in Old French, and second, to the status of verbal prefixes as Path encoders in motion
predicates in Old Catalan.
Page 31
31
Verbal prefixes in Old Romance seem to have an aspectual value, and to be able to change the
aspectual properties of the predicate. In this way, they behave like Latin and other s-framed
languages seen above, in that aspect is expressed through a satellite. This has been shown for
10th–13th century French by Dufresne et al. (2000, 2001, 2003). In (64), for instance, we find
prefix a- with an inchoative meaning, which turns the activity penser ‘think’, into the
achievement s’apenser ‘realise’:
(64) Lors s’=a-pensa Eve qu’=ele n’=avoi huche.
then REFL=at-think.PRF.3SG Eve that=she not=have.IPFV.3SG coffer
‘Then Eve realised that she didn’t have any coffer.’
(Queste du Graal, 116, in Dufresne et al. [2001: 41])
On the other hand, in (65) we find a- (in its older variant ad-) with a completive meaning,
similar to the one shown by con- in (66):
(65) Ad-emplir voeill vostre cumandement.
at-fill.INF want.PRS.1SG your command
‘I want to carry out your orders’
(Roland, XXII, 30, in Dufresne et al. [2003: 37])
(66) Et con-pissa toz mes loviaus.
and with-urinate.PRF.3SG all.PL my.PL wolf_cub.PL
‘And he urinated on all my wolf cubs.’
(Le roman de Renart, Branche I, v. 37, in Dufresne et al. [2003: 37])
Page 32
32
Moreover, according to Dufresne et al. (2003), Old French possessed satellite particles, both
with spatial and aspectual meaning. An example of the latter type is avant, illustrated in (67),
which is strikingly parallel to Latin porro seen above:
(67) Parler avant.
talk.INF forth
‘Talk on.’
(Dufresne et al. 2003: 53)
The second residue of s-framedness in Old Romance that we will look at is the existence, in
Old Catalan, of CDMCs showing a manner conflation, as the ones seen above for Latin and
other s-framed languages. They involve a manner of motion verb and a prefix expressing
directionality, and they contrast with their unprefixed counterparts in that the latter convey
mere activity and not goal-oriented motion. This contrast can be shown in examples (68) and
(69), excerpted from Bartra and Mateu (2005: 93–102):
(68) Córrer ‘run’ vs. acórrer ‘run to’
a. Con vens axi corren (...)?
how come.2SG thus run.GER
‘Why do you come up running?’
b. Venen per a-córrer a les Celles.
come.3PL for to-run.INF to the.F.PL Celles
‘They come to run to the Celles.’
Page 33
33
(69) Caminar ‘walk’ vs. acaminar ‘walk to’
a. Caminar ab calor [...] per a quisvulla és danós.
walk.INF with heat for to anybody is harmful
‘To walk in the heat is harmful for anybody.’
b. Pot hom a-caminar sens peccat a guanyar perdonances.
can.3SG man to-walk.INF without sin to gain forgiveness.PL
‘One can go without sins to obtain forgiveness.’
On the other hand, a proof that the manner conflation pattern is effectively found in Old
Catalan is provided by verbs which may either appear as unergative or unaccusative, as
hallmarked by the choice of auxiliary verb in the perfect tenses, much as happens in Dutch
and German: ésser ‘be’ for the unaccusative variant, as in (70a) and haver ‘have’, for the
unergative variant, as in (70b). Observe, importantly, that both examples are taken from the
same work.
(70) Old Catalan anar ‘go, march’
a. A Roma era anat.
to Rome be.IPFV.3SG march.PTCP.PST.M.SG
‘He had gone to Rome.’
(Vides: 98)
b. Co [...] agés ja anat per III dies.
when have.IPFV.SBJV.3SG already march.PST.PART.M.SG for 3 day.PL
‘And when he had already marched for three days.’
(Vides: 292)
Page 34
34
Variable behaviour verbs involve, in their unaccusative use, a CDMC, as the ones seen above
for Latin, English or Russian, since they express a transition plus a manner of motion. Thus,
in (70a) the verb anar is the result of adjoining a manner root to a light verb v, which takes as
its complement a PathP including the PP a Roma. This analysis is shown in (71):
(71) A Roma era anat.
Complementarily, the predicate in (70b), whose unergative status is further born evidence to
by its licensing the durative adverbial per III dies, demonstrates that this verb had not yet
assumed the path component it has in modern Catalan or modern Italian (andare), and that it
should rather be rendered as ‘march’, or ‘walk’ as its Latin source ambulo, or its Spanish
cognate andar, both meaning ‘walk’. Further evidence of the manner nature of Old Catalan
anar is furnished by the example in (72), where anar is coordinated with a verb clearly
expressing manner, jugar ‘play’:
Page 35
35
(72) Tingué [...] son fill [...] fins a tant que pogué anar e
have.PST.3SG her.M.SG son till to much that can.PRF.3SG walk.INF and
jugar.
play.INF
‘She kept her son until he could walk and play.’
(A-M: Llull Blanq. 2, 5, in Acedo-Matellán and Mateu [2008])
Finally, the alternation shown in (70) is comparable with the well-known ones displayed by
manner of motion verbs in Dutch, as illustrated in (73) through examples by Hoekstra and
Mulder (1990: 4):
(73) a. Jan heft gesprongen.
Jan has jumped
‘Jan has jumped.’
b. Jan is in de sloot gesprongen.
Jan is in the ditch jumped
‘Jan has jumped in the ditch.’
4.2. Prefixed verbs in Romance: a special form of v-framedness
Here we deal with a kind of predicates which have been claimed to involve an s-framed
pattern: Romance prefixed verbs of change of state/location. We will show that they lack,
however, the manner conflation mechanism typical of s-framed languages, alongside other
properties shown by Latin prefixed verbs of the type seen up to now. In fact, we propose
(along lines put forth in Mateu & Rigau 2009), that these Romance predicates represent a v-
framed strategy where the Place component is semantically and phonologically specified.
Page 36
36
The verbs we refer to are represented in all Romance languages that we know of (see Acedo-
Matellán [2006a: 2]); here we give a sample of modern Catalan (from Acedo-Matellán
[2006b: 44]):
(74) Aquest xampú m’=ha a-llisat els cabells.
this shampoo me=has at-smooth.PTCP.PST the.PL hair.PL
‘This shampoo has smoothed my hair.’
(Cf. llis ‘smooth’)
(75) El pas del temps ha a-vinagrat el vi.
the pass of.the time has at-vinegar.PTCP.PST the wine
‘Time has made the wine become like vinegar.’
(Cf. vinagre ‘vinegar’)
(76) L’=Elna ha en-sellat el cavall.
the=Elna has in-saddle.PTCP.PST the horse
‘Elna has saddled the horse.’
(Cf. sella ‘saddle’)
(77) El vent del vespre ha es-boirat el dia.
the wind of.the evening has out-fog.PTCP.PST the day
‘The evening wind has cleared the day from fog.’
(Cf. boira ‘fog’)
(78) La infermera ha as-segut el pacient.
the nurse has at-sit.PTCP.PST the patient
‘The nurse has sat the patient down.’
(Cf. seure ‘sit’)
Page 37
37
As shown by Acedo-Matellán (2006b), what all these verbs have in common is that the prefix
they are provided with furnishes further information on the final state or location involved in
the event and codified by the root. Thus, for instance, in (74) the prefix a- indicates that the
entity represented by els cabells, ‘the hair’, enters the state represented by the root √llis,
‘smooth’, while the prefix es- in (77), on the contrary, expresses that the final state involved is
that of being free from fog (√boir). This specifying ability of the prefix is particularly evident
in the case of minimal pairs such as the one illustrated in (79):
(79) a. L’=absència de vent ha em-boirat el dia.
the=absence of wind has in-fog.PTCP.PST the day
‘The absence of wind has made the day foggy.’
b. El vent ha es-boirat el dia.
the wind has out-fog.PTCP.PST the day
‘The wind has freed the day from fog.’
Both verbs in (79) contain the root of the noun boira ‘fog’, but the former presents an en-
prefix indicating entrance into a state (encoded in the root √boir), while the latter presents an
es- prefix, indicating exit from the same state. These predicates are at first glance alike to the
Latin ones we have examined before, since the prefix is also a specification of a final result
(state or location). Accordingly, some authors have analysed them as s-framed constructions
(e.g., Kopecka 2006). However, what is special about them is the fact that their verb
lexicalises not a Manner component, as is the case with the Latin prefixed predicates we saw
in Section 3, but the Ground component, a lexicalisation pattern which Talmy (2000: 60–61)
claims to be quite rare among the languages of the world (but see Acedo-Matellán [2006a]
Page 38
38
and Kopecka [2006] for qualification). Thus, for instance, -llis- in (74), in representing the
final state of els cabells ‘the hair’, is the abstract Ground in the event of change of state.6 The
analysis we propose for all these verbs is, therefore, the one in (80):
(80) Aquest xampú m’ha allisat els cabells.
As seen in this analysis, the structure we propose for these verbs is non-trivially different
from the one in Latin manner conflation predicates we saw in Section 3. In this case the root
is a Ground, and is no more adjoined to v as a manner conveyor. PathP, thus, contains no
more a full DP in Ground position. As a result, one of the traditionally observed properties of
Latin prefixes, namely, the government it may exert onto an independent DP of the predicate
(Lehmann 1983), is no longer a fact in the new predicates. In the next Latin example for
instance, the prefix ad- governs the dative portis:
(81) Qui ubi *(ad-)equitauit portis.
who.NOM.M.SG as_soon_as at-ride.PRF.3SG door.DAT.PL
‘As soon as he rode up to the doors.’
(Liv. 22, 42, 5)
Page 39
39
The root √llis in (80) is first conflated into the phonologically defective matrix of Place,
which consists of what is overtly manifested as the prefix (a-). This phonological material
conflates up into Path and thence into v, as in any other v-framed structure. Thus, Romance
prefixed predicates, despite apperances, turn out to be analysable as a kind of v-framed
constructions where, unlike simple change of state predicates like melt, the result of the event
is phonologically and semantically split into a Place component (the prefix) and a Ground
component (the verbal root).
Strikingly enough, prefixed Ground-conflating verbs occur already in Classical Latin, as
illustrated by the examples from Crocco Galèas and Iacobini (1993: 55–56) in (82), where the
related noun or adjective is indicated in brackets:
(82) *(ex)pectoro ‘expectorate’ (pectus ‘breast’), *(ir)retio ‘net’ (rete ‘net’), *(in)unco ‘hook’
(uncus ‘hook’), *(e)limo ‘clean’ (limus ‘mud’).
According to Crocco Galèas and Iacobini (1993), however, these verbs increase in number in
the later stages of the language, that is, from the third century AD on, and generalise greatly
in Romance. This increase is to be interpreted, in our opinion, as a signal of the typological
change which transformed Latin into its daughter languages: the systemic s-framed > v-
framed change reinforced the irretio type illustrated in 0, since the prefixes, losing their
original locative value, were desemantised into mere markers of change of state or location.
As argued by Acedo-Matellán (2006a), the semantic weakening of Latin spatial prefixes into
Romance change of state prefixes is related to a change in the nature of the object which
appears in Ground position: when the prefix is semantically strong, a full DP appears in that
position; when it is semantically weak, it is a root what appears as Ground. Thus, the
Page 40
40
phonological and semantic bleaching of Latin prefixes, which is also related to the typological
shift from an s-framed system into a v-framed system, led to the creation of a new type of v-
framed predicates where the Place component is semantically and phonologically specified.
Last, we deal with a group of prefixed Ground-conflating verbs which are often referred to as
deverbal, because they seem to be derived from a verb. An example is asseure ‘sit someone
down’, exemplified in (83):
(83) La infermera ha as-segut el pacient.
the nurse has at-sit.PTCP.PST the patient
‘The nurse has sat the patient down.’
(Cf. seure ‘sit’)
As is the case with the denominal and deadjectival Ground-conflating verbs we saw above,
verbs such as asseure head change-of-state predicates where the object represents the entity
undergoing the change of state/location and the subject is a causative agent. This leads us to
propose an analysis of asseure-predicates as the one in (84):
(84) La infermera ha assegut el pacient.
Page 41
41
Interestingly, the Latin cognates of these verbs show a strikingly different syntax, as gathered
from the next example of adsideo ‘sit next to’:
(85) Hiempsal dextra Adherbalem ad-sedit.
Hiempsal.NOM to_the_right Adherbal.ACC at-sit.PRF.3SG
‘Hiempsal sat down beside Adherbal, on his right.’
(Sal. Jug. 11, 3)
In (85) the subject Hiempsal represents the person sitting down, while the accusative
Adherbalem is the Ground of the sitting event. This locative relation is expressed through the
preverb ad- ‘at’. (85) involves the familiar manner conflation pattern of s-framed languages,
as displayed in (86):
(86) Hiempsal Adherbalem adsedit.
In (86), the root √sed is adjoined to the light verb v, indicating manner. The preposition ad,
which is first merged as Place, taking the full DP Adherbalem as complement, conflates up
into Path above and gets prefixed onto the verb from this position.
Page 42
42
Our hypothesis is that the change from an s-framed system into a v-framed one made it
impossible for predicates such as adsideo in (85) to keep being analysed as manner conflation
structures. As a result, the root was reanalysed as being located in Ground position, selected
by the already impoverished prefix, yielding the asseure-pattern of (83). That asseure-verbs
are reanalyses of old manner conflation structures is suggested by the fact that, as pointed out
by Acedo-Matellán (2006a), these verbs are remarkably scarce in Romance languages. In
Catalan, for instance, only the following five are found:
(87) acréixer ‘increase’ (cf. créixer ‘grow up’), adormir ‘bring to sleep’ (cf. dormir ‘sleep’),
ajeure ‘lie someone down’ (cf. jeure ‘lie’), apujar ‘bring up’ (cf. pujar ‘go up’), asseure ‘sit
someone down’ (cf. seure ‘sit’).
In conclusion, the change from an s- to a v-framed system took place in a rather stepwise
fashion, as the evidence considered in this section shows: the older stages of Romance
admitted s-framed patterns with manner conflation which are lost in the modern languages
(Dufresne et al. 2000, 2001, 2003; Bartra and Mateu 2005). On the other hand, the prefixed v-
framed pattern of Ground conflation represented by Romance verbs like allisar, while being
possible in Latin, constitutes a hallmark of the new system, since the prefixes, losing their
adpositional nature as satellites, were reused as change-of-state specifiers. The gradual change
we have tried to illustrate is in conformity with the fact that, at least as far as the Talmian
typology is concerned, typological groups do not seem to be monolithical classes. In short,
“typologies leak”, and the typological change from s-framed Latin to v-framed Romance
illustrates the diachronic expression of that principle.
Page 43
43
5. Conclusions
In this final section we summarize some relevant conclusions worth being drawn from the
present paper. We have argued that Talmy’s descriptive insight that the motion verb may
generally encode (pure) Manner and Path but not simultaneously can be understood as a
syntactic constraint (Mateu 2002). Furthermore, Latin has been shown to be an s-framed
language, on the basis of the availability of constructions where the Path is realised
independently from the verb: Unselected Object Constructions, Telic Complex Directed
Motion Constructions, Locative Alternation Constructions involving manner conflation and
constructions involving aspectual satellites (Acedo-Matellán, in preparation). In contrast,
modern Romance, particularly modern Catalan, has been shown to be a v-framed language, in
that the Path is typically conflated into the verb and the manner conflation pattern involved in
the abovementioned Latin constructions, is not available. Finally, the change from the Latin s-
framed system to the Romance v-framed system has been characterised through two
phenomena. On the one hand, the oldest stages of Romance did allow the manner conflation
mechanism and, most of all, the expression of the Path as a satellite in a much more
productive way than the modern stages (Dufresne et al. 2000, 2001, 2003; Bartra and Mateu
2005; Acedo-Matellán and Mateu 2008). On the other hand, we have concentrated on the
change from Latin DP-selecting prefixes to Romance root-selecting ones, and we have related
it to the change from a predominantly manner-conflation system to a ground-conflation
system (Acedo-Matellán 2006a).
Page 44
44
References
Acedo-Matellán, Víctor. 2006a. Prefixes in Latin and Romance and the satellite- vs. verb-
framed distinction. In Universitat de Barcelona (ed.), Actes del VII Congrés de
Lingüística General, CD-ROM, Barcelona, Universitat de Barcelona.
Acedo-Matellán, Víctor. 2006b. Una aproximació sintàctica als verbs prefixats en català,
Estudios Catalanes, 4. 41-78.
Acedo-Matellán, Víctor (in preparation). Argument structure in Latin. Barcelona: Universitat
de Barcelona dissertation.
Acedo-Matellán, Víctor & Jaume Mateu. 2008. The path from satellite-framed Indo-European
to verb-framed Romance: A lexical-syntactic account. Paper presented at the 10th
Diachronic Issues in Generative Syntax conference (DIGS X), Cornell University,
Ithaca, 7-9 August.
Ackerman, Farrell. 1992. Complex predicates and morpholexical relatedness: Locative
alternation in Hungarian. In Ivan Sag & Anna Szabolcsi (eds.), Lexical matters, 55-83.
Stanford: CSLI Publications.
Aske, Jon. 1989. Path predicates in English and Spanish: A closer look. In Kira Hall, Michael
Meacham & Richard Shapiro (eds.), Proceedings of the Fifteenth Annual Meeting of
the Berkeley Linguistics Society, 1-14. University of California at Berkeley, CA:
Berkeley Linguistics Society, Inc.
Bartra, Anna & Jaume Mateu. 2005. Aspecte i prefixació verbal en català antic. Caplletra 39.
85-108.
Clark, Eve V. & Herbert H. Clark. 1979. When nouns surface as verbs. Language 55. 767–
811.
Crocco Galèas, Grazia & Claudio Iacobini. 1993. Lo sviluppo del tipo verbale parasintetico in
Latino: i prefissi AD-, IN-, EX-. Quaderni Patavini di Linguistica 12. 31-68.
Page 45
45
Di Sciullo, Anna Maria. 1997. Prefixed-verbs and adjunct identification. In Anna Maria Di
Sciullo (ed.), Projections and interface conditions. Essays on modularity, 52-73. New
York: Oxford University Press.
Dufresne, Monique, Fernande Dupuis & Catherine-Marie Longtin. 2001. Un changement
dans la diachronie du français: la perte de la préfixation aspectuelle en -a. Revue
Québécoise de Linguistique 30. 33-54.
Dufresne, Monique, Fernande Dupuis & Mireille Tremblay. 2000. The role of syntactic
features in historical change. In Steven N. Dworkin & Dieter Wanner (eds.), New
approaches to old problems. Issues in Romance linguistics, 129-149. Amsterdam:
John Benjamins.
Dufresne, Monique, Fernande Dupuis & Mireille Tremblay. 2003. Preverbs and particles in
Old French. In Geert Booij & Jaap van Marle (eds.), Yearbook of morphology, 30-60.
Kluwer.
Folli, Raffaella & Gillian Ramchand. 2005. Prepositions and results in Italian and English: An
analysis from event decomposition. In Henk Verkuyl, Henriette de Swart & Angeliek
van Hout (eds.), Perspectives on aspect, 81-105. Dordrecht: Springer.
García Hernández, Benjamín. 1980. Semántica estructural y lexemática del verbo. Tarragona:
Avesta.
Gehrke, Berit. 2008. Ps in Motion: On the semantics and syntax of P elements and motion
events. Utrecht: Utrecht University dissertation.
Gruber, Jeffrey S. 1965. Studies in lexical relations. Cambridge, MA: MIT dissertation.
Hale, Ken L. & Samuel Jay Keyser. 1993. On argument structure and the lexical expression
of syntactic relations. In Ken L. Hale & Samuel Jay Keyser (eds.), A view from
Building 20: Essays in linguistics in honor of Sylvain Bromberger, 53-109.
Cambridge, MA: MIT Press.
Page 46
46
Hale, Ken L. & Samuel Jay Keyser. 2002. Prolegomenon to a theory of argument structure.
Cambridge, MA: MIT Press.
Hoekstra, Teun. 1988. Small clause results. Lingua 74. 101-139.
Hoekstra, Teun & René Mulder. 1990. Unergatives as copular verbs: Locational and
existential predication. The Linguistic Review 7. 1-79.
Jackendoff, Ray. 1983. Semantics and cognition. Cambridge: MIT Press.
Kopecka, Anetta. 2006. The semantic structure of motion verbs in French: Typological
perspectives. In Maya Hickmann & Stéphane Robert (eds.), Space in languages:
Linguistic systems and cognitive categories, 83-101. Amsterdam: John Benjamins.
Kratzer, Angelika. 2005. Building resultatives. In Claudia Maienborn & Angelika Wöllstein-
Leisten (eds.), Events in syntax, semantics, and discourse, 177-212. Tübingen:
Niemeyer.
Lehmann, Christian. 1983. Latin preverbs and cases. In Harm Pinkster (ed.), Latin linguistics
and linguistic theory, 145-161. Amsterdam: John Benjamins.
Lemaire, Maryvonne. 1983. Un aspect de l’échange des arguments du verbe: les verbes à
double régime de l’objet en latin. Bulletin de la Societé de Linguistique de Paris 78
(1). 283-324.
Levin, Beth & Peter Sells. 2009. Unpredicated particles. In Lian Hee Wee & Linda Uyechi
(ed.), Reality exploration and discovery: Pattern interaction in language and life, 303-
324. Stanford, CA: CSLI Publications.
Mateu, Jaume. 2001. Unselected objects. In Nicole Dehé & Anja Wanner (eds.), Structural
aspects of semantically complex verbs, 83-104. Frankfurt am Main: Peter Lang.
Mateu, Jaume. 2002. Argument structure. Relational construal at the syntax-semantics
interface. Bellaterra: Universitat Autònoma de Barcelona dissertation. Downloadable
at http://seneca.uab.cat/clt/publicacions/tesis/index.html
Page 47
47
Mateu, Jaume. 2005. Arguing our way to the Direct Object Restriction on English
resultatives. Journal of Comparative Germanic Linguistics 8. 55-82.
Mateu, Jaume. 2008. On the l-syntax of directionality/resultativity: The case of Germanic
preverbs. In Anna Asbury, Jakub Dotlačil, Berit Gehrke & Rick Nouwen (eds.),
Syntax and semantics of spatial P, 221-250. Amsterdam: John Benjamins.
Mateu, Jaume & Gemma Rigau. 2002. A minimalist account of conflation processes:
parametric variation at the lexicon-syntax interface. In Artemis Alexiadou (ed.),
Theoretical approaches to universals, 211-236. Amsterdam: John Benjamins.
Mateu, Jaume & Gemma Rigau. 2009. Verb-particle constructions in Romance: a lexical-
syntactic account. Manuscript UAB, Barcelona.
McIntyre, Andrew. 2002. Idiosyncrasy in particle verbs. In Nicole Dehé, Ray Jackendoff,
Andrew McIntyre & Silke Urban (eds.), Verb-particle explorations, 95-118. Berlin &
New York: Mouton de Gruyter.
McIntyre, Andrew. 2004. Event paths, conflation, argument structure, and VP shells.
Linguistics 42 (3). 523-571.
Mulder, René. 1992. The aspectual nature of syntactic complementation. Universiteit Leiden:
HIL dissertations (3).
Ramchand, Gillian & Peter Svenonius. 2002. The lexical syntax and lexical semantics of the
verb-particle construction. In Line Mikkelsen & Christopher Potts (eds.), Proceedings
of the 21 Weast Coast Conference on Formal Linguistics, 101-114. Somerville, MA:
Cascadilla Press.
Spencer, Andrew & Marina Zaretskaya. 1998. Verb prefixation in Russian as lexical
subordination. Linguistics 36. 1-39.
Snyder, William. 2001. On the nature of syntactic variation: evidence from complex
predicates and complex word-formation. Language 77 (2). 324-342.
Page 48
48
Stiebels, Barbara. 1998. Complex denominal verbs in German and the morphology-semantics
interface. In Geert Booij & Jaap van Marle (eds.), Yearbook of Morphology 1997,
265-302. Dordrecht: Kluwer.
Svenonius, Peter. 2003. Limits on P: Filling in holes vs. falling in holes. Nordlyd 31 (2). 431-
445.
Svenonius, Peter. 2004. Slavic prefixes inside and outside VP. Nordlyd 32 (2). 205-253.
Talmy, Leonard. 1985. Lexicalization patterns: Semantic structure in lexical forms. In
Timothy Shopen (ed.), Language typology and syntactic description (vol. 3), 57-149.
Cambridge: Cambridge University Press.
Talmy, Leonard. 1991. Path to realization: A typology of event conflation. In Laurel A.
Sutton, Christopher Johnson & Ruth Shields (eds.), Proceedings of the 17th Annual
Meeting of the Berkeley Linguistics Society, 480-519. University of California at
Berkeley, CA: Berkeley Linguistics Society, Inc.
Talmy, Leonard. 2000. Toward a cognitive semantics. Vol. 2: Typology and process in
concept structuring. Cambridge, MA: MIT Press.
Tungseth, Mai. 2003. Two structural positions for locative and directional PPs in Norwegian
motion constructions. Nordlyd 31 (2). 473-487.
Zubizarreta, María Luisa & Eunjeong Oh. 2007. On the syntactic composition of manner and
motion. Cambridge, MA: MIT Press.
1 If not otherwise indicated, the Latin examples have been extracted from the Latin language
corpus of the Bibliotheca Teubneriana Latina, in its (2002) edition (BTL2). The references of
the works excerpted are also the ones provided in the Bibliotheca.
2 When dealing with apparent counterexamples to Talmy’s typology, Mateu & Rigau (2009)
show that Italian examples like the one in (ia) do not involve the Talmian Co-event pattern
depicted in (4), which they argue holds only for pure manner verbs in path constructions. As
Page 49
49
emphasized by Mateu & Rigau (2009), in Italian the list of verbs that enter into the verb-
particle construction or, more generally, into telic path of motion constructions is limited to
verbs encoding or involving Path: otherwise, they are excluded (see [ib]). In striking contrast
to that, such a restriction does not hold in Germanic (e.g., see [6a-b]).
(i) a. Gianni è corso {via/alla stanza}
Gianni is run.PTCP.M.SG away/to the room
‘Gianni ran away/to the room.’
b. *Gianni è ballato {via/alla stanza}
Gianni is dance.PTCP.M.SG away/to the room
‘Gianni danced away/to the room.’
3 For more examples, see Lemaire (1983).
4 See also Levin and Sells (2007) for relevant observations about the semantic constraints on
these Figure-less predicates.
5 For an extensive semantic description of Latin prefixes, in particular with respect to their
aspectual value, see García Hernández (1980).
6 At first glance sella in (76) is a Figure, rather than a Ground, since it ends up placed onto the
horse. An important qualification to be made here is that we are dealing not with conceptual
scenes, but rather with linguistic representations. Linguistically speaking, sella occupies the
Ground position (the complement of Place): as pointed out by Hale & Keyser (2002), an
appropriate linguistic paraphrase of She saddled the horse would roughly be She provided the
horse with a saddle rather than She put a saddle on the horse. See their work for further
discussion on the l-syntax of locatum verbs.